Economics as science (sigh) again

Economists, as a rule, are highly defensive about being "scientists". By this, they don't mean that they are part of the general social sciences set - a subset of sociology, in fact. No, they mean that they are like physics.
They aren't at all like physics, of course, starting at the ground level. While physicists can start with atoms because atoms are entirely defined by mass and motion, economics has no equivalent. Individuals are not simply defined by mass and motion. Economists have attempted to define them as the atoms of economics, but the arguments for this range from poor to unbelievable. This is even conceded, and got over by pretending that the motion of the individual is wholly defined by the desire to get more. 
Since the very meaning of getting more is not really definable without the system of values defining more, there's really no atomic level to work up from. 
But economists still somehow consider that, since they do hard mathematical work, they must be scientists. And since, at a certain point as undergraduates, they read Friedman's methodological paper about prediction, that science is defined by predicting things. 
So, lately, we've had a round of economists bitching that non-experts are pitching into macro-economics. It started with Scott Sumner here, went to Noah Smith there, and now I'm going to talk about it.

I get tired of the idea that science is entirely defined by the ability to predict things, which is like defining a car as the ability to go 70 miles per hour. Many other things than cars possess this ability, from hurricanes to peregrine falcons. But it is into the prediction hole that the whole incredibly badly formed debate around economics turns. Did economists predict the 2008 downturn or not? and then we are off to the races. 
But I'll bite on this for at least a second and ask what I think are more pertinent questions. 
-- what economist in the seventies or eighties predicted that the medium American wage would effectively stagnate for the next thirty or more years? 
-Which predicted that household debt would begin to equal household wealth for the medium household? 
-Who predicted that, even with the advent of IRAs, mutual funds, and 401ks, the shape of the ownership of all financial instruments would essentially remain the same in 2014 as it was in 1979? 
- Who predicted that the last American trade surplus would be in 1976?
My questions are all invidious. They are all about trends. The prediction biz, as defined by economists, is a pretty narrow thing about particular events. The local downfall, the inflation figure for the next quarter. But it is long term  long term trends, which are the meat and drink of the prediction biz of the natural sciences. The prediction of the course of a single atom isn't in it. The trend, going back billions of years, is. 
What seems evident to me is that economics, as a branch of sociology, can produce ideal models of various economies (not just capitalist ones) and capture broad trends within them. But it isn't very good, even so, at predicting long term trends at any middle distance - and as for up close, no way. Marx's prediction of the inevitable fall of the rate of profit, founded on classical economics, is a good instance of the use of trends - the ideal model of capitalism he constructed would seem to require it. One can see how the physical limitations inhering on labor time would even make it, at some point, true. But it has turned out to be only a factor, and a reversible one, in capitalist business cycles. Or perhaps I should say, a factor with varying weight.